Both Claude Code and GitHub Copilot CLI now support remote control of agentic coding sessions from a phone or browser, letting you keep execution local while steering the session remotely. Claude Code uses `/remote-control` (or `/rc`) and connects via claude.ai/code or the Claude app, with an outbound-only HTTPS architecture. GitHub Copilot CLI uses `/remote on` and connects via GitHub.com or GitHub Mobile. Key differences: Copilot CLI has a built-in `/keep-alive` command with configurable durations, while Claude Code offers a server mode with per-session git worktrees for parallel sessions. Both tools print a session URL and support QR codes for quick mobile access.
Table of contents
Claude Code: /remote-controlGitHub Copilot CLI: /remoteSide by sideMore informationQuestions this post answers
How do I control my Claude Code session from my phone while away from my desk?
Enable remote control inside a running session with `/remote-control` (or `/rc`), or start one with `claude --remote-control`. Claude Code prints a session URL and optionally a QR code that connects to claude.ai/code or the Claude app. The local process makes an outbound HTTPS connection to Anthropic's API — no inbound port is opened. Your phone renders the conversation and sends prompts; execution and file access stay on your machine. Developers managing agentic Claude Code sessions on the go track tips like these on daily.dev.
How does GitHub Copilot CLI remote control differ from regular session syncing?
Regular Copilot CLI sessions sync to your GitHub account and appear as view-only on GitHub.com and GitHub Mobile — you can read output but cannot send prompts. Only sessions explicitly started with `/remote on` or `--remote` (or `"remoteSessions": true` in settings.json) are steerable. The `/keep-alive` command (with options like `busy` or a fixed duration like `8h`) also prevents your laptop from sleeping and killing the session. Developers choosing between Copilot CLI and Claude Code for remote agentic work find comparisons like this on daily.dev.
How do I run parallel Claude Code agentic sessions without file conflicts?
Use Claude Code's server mode with `claude remote-control --spawn worktree`, which gives each on-demand session its own git worktree so parallel sessions don't compete over the same files. Sessions appear as separate chat threads in the Claude app, with push notifications when one finishes or needs a decision. Teams running multiple concurrent AI coding agents can follow patterns like this on daily.dev.